AI Act. Obligations and Deadlines after the Omnibus (2026 Status)

Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence (AI Act), as amended by Regulation (EU) 2026/1744 (Digital Omnibus on AI)

In force EU-wide
StatusIn forceApplies since 01/08/2024
Legal actEU Regulationdirectly applicable
Fines up to7%details below

Overview

The EU AI Act (Regulation (EU) 2024/1689) is the first comprehensive law regulating artificial intelligence. It follows a risk-based approach: the higher the risk an AI system poses to health, safety and fundamental rights, the stricter the obligations. As a regulation, it applies directly in all member states.

In 2026 the AI Act was amended by the Digital Omnibus on AI (Regulation (EU) 2026/1744), which entered into force on 27 July 2026. Above all, it postponed the deadlines for high-risk systems and softened the AI literacy obligation.

Timeline

DateWhat applies
01.08.2024Entry into force
02.02.2025Prohibited AI practices and AI literacy measures
02.08.2025Obligations for general-purpose AI (GPAI) models, governance, penalties
02.08.2026Transparency obligations under Art. 50, Commission enforcement powers over GPAI providers
02.12.2026Machine-readable marking of AI-generated content also for systems placed on the market before 02.08.2026
02.08.2027GPAI models placed on the market before 02.08.2025; at least one AI regulatory sandbox per member state
02.12.2027High-risk AI systems under Annex III (previously 02.08.2026)
02.08.2028High-risk AI in Annex I products (previously 02.08.2027)

The Four Risk Levels

Prohibited AI Practices (since 02.02.2025)

  • Manipulative or deceptive techniques that materially distort behaviour and cause harm
  • Exploiting vulnerabilities due to age, disability or social situation
  • Social scoring – by public and private actors
  • Predicting criminal offences based solely on profiling
  • Untargeted scraping of facial images from the internet or CCTV
  • Emotion recognition in the workplace and in education (except for medical or safety reasons)
  • Biometric categorisation by sensitive characteristics such as religion or sexual orientation
  • Real-time remote biometric identification in public spaces for law enforcement (with narrow exceptions)

The 2026 Omnibus added AI systems that generate non-consensual intimate imagery (“nudifier” apps) or child sexual abuse material to the list.

High-Risk AI Systems

Two groups are high-risk:

  • Annex I: AI as a safety component of products covered by EU product legislation, such as machinery, medical devices, toys or lifts – obligations from 02.08.2028
  • Annex III: AI in specific use cases – obligations from 02.12.2027:
    • Biometrics
    • Critical infrastructure
    • Education and vocational training (e.g. admission, grading)
    • Employment and workforce management (e.g. recruitment, promotion, performance evaluation)
    • Access to essential services (e.g. creditworthiness, pricing in life and health insurance, public benefits)
    • Law enforcement, migration, asylum and border control
    • Administration of justice and democratic processes

Limited Risk: Transparency Obligations (since 02.08.2026)

  • People must be able to recognise that they are interacting with an AI system (e.g. chatbots)
  • AI-generated audio, image, video and text content must be marked in a machine-readable way
  • Deepfakes must be disclosed as artificially generated
  • Deployers of emotion recognition or biometric categorisation must inform the persons concerned

Minimal Risk

Most AI applications – such as spam filters, recommender systems or AI in video games – have no specific obligations apart from AI literacy measures.

Obligations by Role

RoleKey obligations for high-risk AI
Provider (develops or has developed and places on the market under its own name)Risk management, data quality, technical documentation, logging, human oversight, accuracy and cybersecurity, quality management, conformity assessment, CE marking, registration in the EU database
Deployer (uses AI professionally)Use according to instructions, human oversight by competent persons, monitoring, keep logs for at least 6 months, inform employees; fundamental rights impact assessment for public bodies and for credit and insurance decisions
Importer and distributorCheck that conformity assessment, CE marking and documentation are in place

Anyone who substantially modifies a high-risk system or markets it under their own name becomes a provider.

AI literacy (Art. 4): Providers and deployers must take measures to ensure their staff have sufficient AI literacy. The Omnibus turned this into an obligation to take measures – a specific level of literacy per person no longer has to be demonstrated.

General-Purpose AI Models (GPAI)

Since 2 August 2025, providers of GPAI models – such as large language models – must draw up technical documentation, inform downstream providers, have a policy to comply with copyright law and publish a summary of the training data. Models with systemic risk are additionally subject to model evaluations, reporting of serious incidents and cybersecurity requirements. The GPAI Code of Practice serves as a way to demonstrate compliance.

Penalties

InfringementFine up to
Prohibited AI practicesEUR 35 m or 7% of worldwide annual turnover
Other obligations (e.g. high-risk, transparency)EUR 15 m or 3%
Incorrect or misleading information to authoritiesEUR 7.5 m or 1%

The higher amount applies; for SMEs and start-ups, the lower one.

Implementation in DACH

Germany: The AI Market Surveillance and Innovation Act (KI-MIG) has been in force since 29 July 2026. The Federal Network Agency is the central market surveillance authority, single point of contact and complaints body; BaFin remains responsible in the financial sector. The Federal Network Agency also runs an AI regulatory sandbox.

Austria: Since 2024, the AI Service Desk at RTR has provided information and advice. The designation of market surveillance authorities had not been completed as of October 2026.

Switzerland: The AI Act does not apply. Switzerland signed the Council of Europe AI Convention in March 2025 and relies on targeted amendments to existing laws rather than a dedicated AI act. Swiss providers fall under the AI Act if they place AI systems on the EU market or their output is used in the EU.

Where AI processes personal data, the GDPR also applies – for example to automated individual decisions. The new Product Liability Directive treats software including AI as a product. For AI in connected products, the Cyber Resilience Act applies as well.

AI Act: Does it affect you?

Find out if and how this regulation affects your company – we're happy to advise you.

Frequently Asked Questions

Who is affected by the AI Act?

Providers placing AI systems on the EU market, deployers using AI systems professionally in the EU, and importers and distributors – regardless of whether the company is based in the EU. Providers outside the EU are also covered if the output of their AI systems is used in the EU.

When does the AI Act apply?

The regulation entered into force on 1 August 2024 and applies in stages: prohibitions and AI literacy since 2 February 2025, obligations for general-purpose AI models since 2 August 2025, transparency obligations since 2 August 2026. After the Digital Omnibus, high-risk obligations apply to Annex III systems from 2 December 2027 and to AI in regulated products from 2 August 2028.

Were the high-risk deadlines postponed?

Yes. The Digital Omnibus on AI, Regulation (EU) 2026/1744, entered into force on 27 July 2026. It moves the obligations for Annex III high-risk systems from 2 August 2026 to 2 December 2027, and for high-risk AI in Annex I products to 2 August 2028.

Is my AI system high-risk?

High-risk systems are mainly AI used as a safety component of regulated products (Annex I) and AI in the areas listed in Annex III – such as recruitment and performance evaluation of employees, creditworthiness assessment, pricing in life and health insurance, education, critical infrastructure, biometrics, law enforcement and migration.

What obligations apply to companies that only use AI?

Deployers must take AI literacy measures for their staff and observe transparency obligations, for example for deepfakes. For high-risk systems they must also use them according to the instructions, ensure human oversight, monitor operation, keep logs for at least six months and inform affected employees.

What are the penalties?

Up to EUR 35 million or 7% of worldwide annual turnover for prohibited practices, up to EUR 15 million or 3% for breaches of other obligations and up to EUR 7.5 million or 1% for supplying incorrect information to authorities. For SMEs the lower amount applies.

Who supervises the AI Act in Germany and Austria?

In Germany the Federal Network Agency is the central market surveillance authority under the KI-MIG, in force since 29 July 2026; BaFin remains responsible in the financial sector. In Austria the AI Service Desk at RTR provides information; market surveillance authorities had not been fully designated as of October 2026.